NYC hospital chief proposes AI-only cancer screening to cut costs, but radiologists warn of patient safety risks as regulatory battles loom.
  • EvilBit@lemmy.worldEnglish
    6·
    10 hours ago

    Exactly. People keep shoehorning Large Language Models into non-linguistic domains, and that’s dangerous. Human language, with respect to the training sets used, is inherently subjective and imperfect. Healthcare is very fault-intolerant.